Identifying Linear Relational Concepts in Large Language Models

arXiv:2311.08968v2 Announce Type: replace
Abstract: Transformer language models (LMs) have been shown to represent concepts as directions in the latent space of hidden activations. However, for any human-interpretable concept, how can we find its direction in the latent space? We present a technique called linear relational concepts (LRC) for finding concept directions corresponding to human-interpretable concepts by first modeling the relation between subject and object as a linear relational embedding (LRE). We find that inverting the LRE and using earlier …

abstract arxiv concept concepts cs.ai cs.cl hidden however human language language models large language large language models linear lms relational space transformer transformer language models type